Ep. 105 -Michael Leggs: The Essay Was Never Evidence of Thinking

What if the real crisis AI has exposed in higher education isn't student dishonesty, but the fact that we've been treating final papers as evidence of thinking—when AI can now simulate that thinking in seconds? In this episode, host Jeff Dillon sits down with Michael Leggs, who has taught English at St. Paul College for 21 years and is the founder of Smartlby, an AI‑powered revision tutor designed to help students strengthen their writing through guided reflection—without writing for them. Michael brings a rare perspective: he's not a technologist or an administrator talking about AI in the abstract. He's the person reading the papers, watching students struggle, and redesigning assignments in real time. Michael argues that detection software is a triage response, not a long‑term solution. The deeper problem is that our curriculum and assessment models were already fragile—AI just made it visible. He explains why surveillance tools create environments of mistrust, why revision should be treated as evidence of thinking, and how tools like Smartlby can guide students through a Socratic dialogue about their own writing without ever rewriting a single sentence. For any faculty member, department chair, or academic leader trying to figure out what actually works in the AI era, this episode offers a grounded, practical, and deeply human perspective.  Key Takeaways  The Real Problem Isn't Cheating—It's Assessment Design: AI didn't create the problem; it exposed a weakness that was already there. We've been treating final papers as evidence of thinking, but AI can now simulate that product. The solution isn't more surveillance—it's redesigning how we assess learning.  Detection Software Is a Triage Response, Not a Solution: Institutions rushed to buy detection tools because it was something they could control. But surveillance creates environments of mistrust—students know we don't trust them, and that damages the learning relationship.  Revision Is Evidence of Thinking: The shift needs to be from product assessment to process assessment. When a student revises, they are thinking. Tools that track revision give instructors visible evidence of that thinking process.  Process Assessment Is Instructionally Scalable—Not Just Technically Scalable: Many tools are technically scalable (they work at scale) but not instructionally scalable (they don't actually help teachers teach better). Smartlby was designed by a teacher with 21 years of classroom experience, so it solves problems faculty actually face.  AI Closes the Production Gap—But the Learning Gap Is Still Open: Students can produce more effectively with AI assistance. But producing more isn't the same as learning more. The goal should be to close the learning gap, not just the production gap.  Redesign Starts at the Grassroots Level: Faculty sharing ideas about how they're redesigning assignments is where real change begins. It doesn't have to be a top‑down mandate. Department‑level collaboration on assignment types can create scalable, shared solutions.  Surveillance Tools Should Be Abolished in Five Years: Michael's hope is that within five years, institutions will stop policing student writing and instead focus on ethical AI use, assistive tools, and redesigning assignments that actually teach students how to think—not just how to produce.   The Signal Newsletter: https://edtechconnect.com/newsletter  <... Chapters (00:00:00) - The Real Problem with AI in Higher Education(00:02:11) - The Name of Smartlby(00:03:02) - What Pulled You Back into the Classroom?(00:04:47) - The Tipping Point for Academic Writing(00:06:44) - SmartAlby on How to Prevent Student Cheating(00:15:41) - Does generative AI Close the Production Gap?(00:21:33) - How to Keep AI Out of the Composition Classroom